Abstract
Many teachers are concerned that students with behavior problems may strain teaching, classmates, or themselves. Although these concerns are understandable, research has yet to clarify the extent to which the teacher-assigned label "behavior problems" is substantiated by students' actual behaviors - and the extent to which it is due to other conditions biasing teachers' perceptions. Addressing this research gap, the present paper explores the conditions of teachers' labeling tendencies in a multi-informant, multilevel survey study. A total of 85 elementary school teachers and 1,412 students answered a questionnaire. Teachers reported the degree to which they consider each student in their class to have behavior problems (labeling tendency). Four randomly assigned classmates rated the frequency of the students' undisciplined behaviors as a presumed condition of teachers' labeling tendencies. Further, we assessed non-behavioral student characteristics, teacher characteristics, and contextual factors as additional labeling conditions. A two-level structural equation model yielded significant effects for the individual students' indiscipline frequency, sex, and subjective experience of instructional clarity on the teachers' tendencies to label them (level 1). At level 2, the teachers' general sensitivity to disturbances and work-related stress experience were found to be significant conditions of their general labeling tendency across students. No significant effects were found for students' collective level of indiscipline or their average experience of instructional clarity in the classes. In sum, the effects explain 54% of the variance in teachers' tendency to label individual students as having behavior problems (level 1) and 37% of the variance in their general labeling tendency across students (level 2). The main findings indicate that the individual students' behaviors largely substantiated the teachers' labeling tendencies (large effect size) - the additional labeling conditions that had little or nothing to do with the students' behavior imply perception biases (small to medium effect sizes).